[EmotionML] another imlementation and questions

Hello all,
my name's Alexandre Denis and I'm working on a Java fully compliant 
EmotionML implementation. I'm now trying to double check everything 
before I could upload the library somewhere. I wanted to know if there 
existed a corpus of badly formatted emotionml files. I created one that 
tests so far most of the assertions listed in the implementation report, 
but maybe there exists one that you are using to validate the existing 
implementations.

Moreover while doing the double check I noticed several problems (or 
misunderstandings from me!) with both the latest schema and assertions:

105: The <emotionml> element MAY contain a single <info> element (and 
155, 601, 606)
- does that mean that we could not have two <info> elements ? I'm not 
sure since the latest schema does not specify that. Maybe the number of 
<info> should be a required assertion rather than an optional one.

215: A <category> MAY contain a <trace> element (and others)
- same question, is it allowed to have several <trace> elements ? 
neither the schema nor the assertions restrict that I think.

410: The <reference> element MUST contain a "uri" attribute
- it is not enforced in the schema, this should be easy to fix

172: The "version" attribute of <emotion>, if present, MUST have the 
value "1.0"
- I think there's an error in the schema, it should not be "optional 
with default value 1.0" but rather "optional with fixed value 1.0"

224: A <dimension> MUST contain either a "value" attribute or a <trace> 
element
216: A <category> MUST not contain both a "value" attribute and a 
<trace> element (and others)
- it is not enforced in the schema, I'm not a schema guru but I think 
it's possible to enforce that

210: If the <category> element is used, a category vocabulary MUST be 
declared using a "category-set" attribute on either the enclosing 
<emotion> element or the root element <emotionml> (and others)
- not enforced by schema, I'm not sure how easy it is to setup

I totally understand that it is not possible to do *all* the 
verifications in the schema. But maybe then the section "4.3 Schema 
validation and processor validation of EmotionML documents" could 
mention these limitations (or even better to list them) since it now 
only refers to the naming validation restrictions.

Actually in the implementation I did it the other way around, implement 
most of the assertions verification without schema validation, then in 
the end added another layer of schema validation.
thanks for your help,
best regards,
Alexandre

Received on Thursday, 18 April 2013 15:38:49 UTC